| Summary: | Traffic demands on optical transport networks continue to grow, both in numbers and in size, at an incredible rate. Consequently, the efficient use of network resources has never been as important as today. A possible solution to this problem is to plan, develop and implement efficient algorithms for static and/or dynamic applications in order to minimize the probability of blocking and/or minimizing the number of wavelengths. Static Routing and Wavelength Assignment (RWA) algorithms use a given set of optical path requests and are intended to provide a long-term plan for future traffic. Static RWA algorithms are important for current and future WDM (Wavelength-Division Multiplexing) networks, especially when there is no wavelength conversion, the network is highly connected or the traffic load is moderate to high. In this dissertation, we propose to develop an optical network planning tool capable of choosing the best optical path and assigning as few wavelengths as possible. This tool is structured in five phases: in the first phase, the network physical topology is defined by the adjacency matrix or by the cost matrix and the logical topology is defined by the traffic matrix; in a second phase, the Dijkstra algorithm is used to find the shortest path for each connection; in the third phase, the traffic routing is accomplished considering one traffic unit between the source and destination nodes; in the fourth phase, the paths are ordered using various ordering strategies, such as Shortest Path First, Longest Path First and Random Path Order; finally, in the fifth phase, the heuristic algorithms for wavelength assignment, such as Graph Coloring, First-Fit and Most-Used are used. This tool is first tested on small networks (e.g. ring and mesh topologies), and then applied to real networks (e.g. COST 239, NSFNET and UBN topologies). We have concluded that the number of wavelengths calculated for each network is almost independent of the Wavelength Assignment (WA) heuristics, as well as the ordering strategy, when a full mesh logical topology is considered.
|
|---|